diff options
Diffstat (limited to 'docs/report/dpdk_performance_tests')
5 files changed, 140 insertions, 268 deletions
diff --git a/docs/report/dpdk_performance_tests/csit_release_notes.rst b/docs/report/dpdk_performance_tests/csit_release_notes.rst index 363c2e70f1..28e6a37768 100644 --- a/docs/report/dpdk_performance_tests/csit_release_notes.rst +++ b/docs/report/dpdk_performance_tests/csit_release_notes.rst @@ -4,7 +4,26 @@ Release Notes Changes in |csit-release| ------------------------- -No code changes apart from bug fixes. +#. **DPDK performance tests** + + - *MRR tests* - Maximum Receive Rate tests measure the packet forwarding rate + under the maximum load offered by traffic generator over a set trial + duration, regardless of packet loss. MRR tests are used for continuous + performance trending and for comparison between releases. + + - *MLR tests* - NDR and PDR tests measure the packet forwarding rate using + MLRsearch library by traffic generator. All tests that previously used + binary search were converted to MLRsearch. + + - *2-node tests* - Set of 2-node tests covering testpmd and l3fwd. + + - Increased coverage of NIC specific tests (Intel-xxv710-da2, Intel-x710) + + - *Generated tests* - Simplified and unified test structure, + semi-autogenerated by generator script. Test generator is currently able + to create test combinations with various frame size and cores combinations. + All existing test cases were converted to new format. + Performance Changes ------------------- @@ -12,19 +31,15 @@ Performance Changes Relative performance changes in measured packet throughput in |csit-release| are calculated against the results from |csit-release-1| report. Listed mean and standard deviation values are computed based on -a series of the same tests executed against respective VPP releases to +a series of the same tests executed against respective DPDK releases to verify test results repeatability, with percentage change calculated for -mean values. Note that the standard deviation is quite high for a small -number of packet throughput tests, what indicates poor test results -repeatability and makes the relative change of mean throughput value not -fully representative for these tests. The root causes behind poor -results repeatability vary between the test cases. +mean values. NDR Changes ~~~~~~~~~~~ -NDR throughput changes between releases are available in a -CSV and pretty ASCII formats: +NDR throughput changes between releases are available in a CSV and pretty ASCII +formats: - `csv format for 1t1c <../_static/dpdk/performance-changes-1t1c-ndr.csv>`_, - `csv format for 2t2c <../_static/dpdk/performance-changes-2t2c-ndr.csv>`_, @@ -36,14 +51,14 @@ CSV and pretty ASCII formats: Test results have been generated by `FD.io test executor dpdk performance job 3n-hsw`_ with Robot Framework result - files csit-vpp-perf-|srelease|-\*.zip + files csit-dpdk-perf-|srelease|-\*.zip `archived here <../_static/archive/>`_. PDR Changes ~~~~~~~~~~~ -PDR throughput changes between releases are available in a -CSV and pretty ASCII formats: +PDR throughput changes between releases are available in a CSV and pretty ASCII +formats: - `csv format for 1t1c <../_static/dpdk/performance-changes-1t1c-pdr.csv>`_, - `csv format for 2t2c <../_static/dpdk/performance-changes-2t2c-pdr.csv>`_, @@ -55,20 +70,20 @@ CSV and pretty ASCII formats: Test results have been generated by `FD.io test executor dpdk performance job 3n-hsw`_ with Robot Framework result - files csit-vpp-perf-|srelease|-\*.zip + files csit-dpdk-perf-|srelease|-\*.zip `archived here <../_static/archive/>`_. Comparison Across Testbeds -------------------------- -Relative performance changes in measured packet throughputon 3-Node Skx testbed +Relative performance changes in measured packet throughput on 3-Node Skx testbed are calculated against the results measured on 3-Node Hsw testbed. NDR Changes ~~~~~~~~~~~ -NDR changes between testbeds are available in a -CSV and pretty ASCII formats: +NDR throughput changes between testbeds are available in a CSV and pretty ASCII +formats: - `csv format for ndr <../_static/dpdk/performance-compare-testbeds-3n-hsw-3n-skx-ndr.csv>`_, - `pretty ASCII format for ndr <../_static/dpdk/performance-compare-testbeds-3n-hsw-3n-skx-ndr.txt>`_. @@ -79,14 +94,14 @@ CSV and pretty ASCII formats: `FD.io test executor dpdk performance job 3n-hsw`_ and `FD.io test executor dpdk performance job 3n-skx`_ with Robot Framework result - files csit-vpp-perf-|srelease|-\*.zip + files csit-dpdk-perf-|srelease|-\*.zip `archived here <../_static/archive/>`_. PDR Changes ~~~~~~~~~~~ -PDR throughput changes between testbeds are available in a -CSV and pretty ASCII formats: +PDR throughput changes between testbeds are available in a CSV and pretty ASCII +formats: - `csv format for pdr <../_static/dpdk/performance-compare-testbeds-3n-hsw-3n-skx-pdr.csv>`_, - `pretty ASCII format for pdr <../_static/dpdk/performance-compare-testbeds-3n-hsw-3n-skx-pdr.txt>`_. @@ -97,7 +112,7 @@ CSV and pretty ASCII formats: `FD.io test executor dpdk performance job 3n-hsw`_ and `FD.io test executor dpdk performance job 3n-skx`_ with Robot Framework result - files csit-vpp-perf-|srelease|-\*.zip + files csit-dpdk-perf-|srelease|-\*.zip `archived here <../_static/archive/>`_. Known Issues @@ -108,10 +123,5 @@ Here is the list of known issues in |csit-release| for Testpmd performance tests +---+---------------------------------------------------+------------+-----------------------------------------------------------------+ | # | Issue | Jira ID | Description | +---+---------------------------------------------------+------------+-----------------------------------------------------------------+ -| 1 | Testpmd in 1t1c and 2t2c setups - large variation | CSIT-569 | Suspected NIC firmware or DPDK driver issue affecting NDR | -| | of discovered NDR throughput values across | | throughput. Applies to XL710 and X710 NICs, no issues observed | -| | multiple test runs with xl710 and x710 NICs. | | on x520 NICs. | -+---+---------------------------------------------------+------------+-----------------------------------------------------------------+ -| 2 | Lower than expected NDR throughput with xl710 | CSIT-571 | Suspected NIC firmware or DPDK driver issue affecting NDR | -| | and x710 NICs, compared to x520 NICs. | | throughput. Applies to XL710 and X710 NICs. | +| | No known issues | | | +---+---------------------------------------------------+------------+-----------------------------------------------------------------+ diff --git a/docs/report/dpdk_performance_tests/overview.rst b/docs/report/dpdk_performance_tests/overview.rst index b38f9595be..e6abb53c90 100644 --- a/docs/report/dpdk_performance_tests/overview.rst +++ b/docs/report/dpdk_performance_tests/overview.rst @@ -1,239 +1,127 @@ Overview ======== -Tested Physical Topologies --------------------------- - -CSIT DPDK performance tests are executed on physical baremetal servers hosted -by :abbr:`LF (Linux Foundation)` FD.io project. Testbed physical topology is -shown in the figure below.:: - - +------------------------+ +------------------------+ - | | | | - | +------------------+ | | +------------------+ | - | | | | | | | | - | | <-----------------> | | - | | DUT1 | | | | DUT2 | | - | +--^---------------+ | | +---------------^--+ | - | | | | | | - | | SUT1 | | SUT2 | | - +------------------------+ +------------------^-----+ - | | - | | - | +-----------+ | - | | | | - +------------------> TG <------------------+ - | | - +-----------+ - -SUT1 and SUT2 are two System Under Test servers (Cisco UCS C240, each with two -Intel XEON CPUs), TG is a Traffic Generator (TG, another Cisco UCS C240, with -two Intel XEON CPUs). SUTs run Testpmd/L3FWD SW SW application in Linux -user-mode as a Device Under Test (DUT). TG runs TRex SW application as a packet -Traffic Generator. Physical connectivity between SUTs and to TG is provided -using different NIC models that need to be tested for performance. Currently -installed and tested NIC models include: - -#. 2port10GE X520-DA2 Intel. -#. 2port10GE X710 Intel. -#. 2port10GE VIC1227 Cisco. -#. 2port40GE VIC1385 Cisco. -#. 2port40GE XL710 Intel. - -From SUT and DUT perspective, all performance tests involve forwarding packets -between two physical Ethernet ports (10GE or 40GE). Due to the number of -listed NIC models tested and available PCI slot capacity in SUT servers, in -all of the above cases both physical ports are located on the same NIC. In -some test cases this results in measured packet throughput being limited not -by VPP DUT but by either the physical interface or the NIC capacity. - -Going forward CSIT project will be looking to add more hardware into FD.io -performance labs to address larger scale multi-interface and multi-NIC -performance testing scenarios. - -Note that reported DUT (DPDK) performance results are specific to the SUTs -tested. Current :abbr:`LF (Linux Foundation)` FD.io SUTs are based on Intel -XEON E5-2699v3 2.3GHz CPUs. SUTs with other CPUs are likely to yield different -results. A good rule of thumb, that can be applied to estimate DPDK packet -thoughput for Phy-to-Phy (NIC-to-NIC, PCI-to-PCI) topology, is to expect -the forwarding performance to be proportional to CPU core frequency, -assuming CPU is the only limiting factor and all other SUT parameters -equivalent to FD.io CSIT environment. The same rule of thumb can be also -applied for Phy-to-VM/LXC-to-Phy (NIC-to-VM/LXC-to-NIC) topology, but due to -much higher dependency on intensive memory operations and sensitivity to Linux -kernel scheduler settings and behaviour, this estimation may not always yield -good enough accuracy. - -For detailed :abbr:`LF (Linux Foundation)` FD.io test bed specification and -physical topology please refer to `LF FD.io CSIT testbed wiki page -<https://wiki.fd.io/view/CSIT/CSIT_LF_testbed>`_. +For description of physical testbeds used for DPDK performance tests +please refer to :ref:`tested_physical_topologies`. -Performance Tests Coverage --------------------------- +.. _tested_logical_topologies: -Performance tests are split into two main categories: +Logical Topologies +------------------ -- Throughput discovery - discovery of packet forwarding rate using binary search - in accordance to :rfc:`2544`. +CSIT DPDK performance tests are executed on physical testbeds described +in :ref:`tested_physical_topologies`. Based on the packet path thru +server SUTs, three distinct logical topology types are used for DPDK DUT +data plane testing: - - NDR - discovery of Non Drop Rate packet throughput, at zero packet loss; - followed by one-way packet latency measurements at 10%, 50% and 100% of - discovered NDR throughput. - - PDR - discovery of Partial Drop Rate, with specified non-zero packet loss - currently set to 0.5%; followed by one-way packet latency measurements at - 100% of discovered PDR throughput. - -- Throughput verification - verification of packet forwarding rate against - previously discovered throughput rate. These tests are currently done against - 0.9 of reference NDR, with reference rates updated periodically. - -|csit-release| includes following performance test suites, listed per NIC type: +#. NIC-to-NIC switching topologies. -- 2port10GE X520-DA2 Intel +NIC-to-NIC Switching +~~~~~~~~~~~~~~~~~~~~ - - **L2IntLoop** - L2 Interface Loop forwarding any Ethernet frames between - two Interfaces. +The simplest logical topology for software data plane application like +DPDK is NIC-to-NIC switching. Tested topologies for 2-Node and 3-Node +testbeds are shown in figures below. -- 2port40GE XL710 Intel +.. only:: latex - - **L2IntLoop** - L2 Interface Loop forwarding any Ethernet frames between - two Interfaces. - -- 2port10GE X520-DA2 Intel - - - **IPv4 Routed Forwarding** - L3 IP forwarding of Ethernet frames between - two Interfaces. - -Execution of performance tests takes time, especially the throughput discovery -tests. Due to limited HW testbed resources available within FD.io labs hosted -by Linux Foundation, the number of tests for NICs other than X520 (a.k.a. -Niantic) has been limited to few baseline tests. Over time we expect the HW -testbed resources to grow, and will be adding complete set of performance -tests for all models of hardware to be executed regularly and(or) -continuously. - -Performance Tests Naming ------------------------- + .. raw:: latex -|csit-release| follows a common structured naming convention for all performance -and system functional tests, introduced in CSIT-17.01. + \begin{figure}[H] + \centering + \includesvg[width=0.90\textwidth]{../_tmp/src/vpp_performance_tests/logical-2n-nic2nic} + \label{fig:logical-2n-nic2nic} + \end{figure} -The naming should be intuitive for majority of the tests. Complete description -of CSIT test naming convention is provided on :ref:`csit_test_naming`. +.. only:: html -Methodology: Multi-Core and Multi-Threading -------------------------------------------- + .. figure:: ../vpp_performance_tests/logical-2n-nic2nic.svg + :alt: logical-2n-nic2nic + :align: center -**Intel Hyper-Threading** - |csit-release| performance tests are executed with -SUT servers' Intel XEON processors configured in Intel Hyper-Threading Disabled -mode (BIOS setting). This is the simplest configuration used to establish -baseline single-thread single-core application packet processing and forwarding -performance. Subsequent releases of CSIT will add performance tests with Intel -Hyper-Threading Enabled (requires BIOS settings change and hard reboot of -server). -**Multi-core Tests** - |csit-release| multi-core tests are executed in the -following VPP thread and core configurations: +.. only:: latex -#. 1t1c - 1 pmd worker thread on 1 CPU physical core. -#. 2t2c - 2 pmd worker threads on 2 CPU physical cores. + .. raw:: latex -Note that in many tests running Testpmd/L3FWD reaches tested NIC I/O bandwidth -or packets-per-second limit. + \begin{figure}[H] + \centering + \includesvg[width=0.90\textwidth]{../_tmp/src/vpp_performance_tests/logical-3n-nic2nic} + \label{fig:logical-3n-nic2nic} + \end{figure} -Methodology: Packet Throughput ------------------------------- +.. only:: html -Following values are measured and reported for packet throughput tests: + .. figure:: ../vpp_performance_tests/logical-3n-nic2nic.svg + :alt: logical-3n-nic2nic + :align: center -- NDR binary search per :rfc:`2544`: +Server Systems Under Test (SUT) runs DPDK Testpmd/L3FWD application in +Linux user-mode as a Device Under Test (DUT). Server Traffic Generator (TG) +runs T-Rex application. Physical connectivity between SUTs and TG is provided +using different drivers and NIC models that need to be tested for performance +(packet/bandwidth throughput and latency). - - Packet rate: "RATE: <aggregate packet rate in packets-per-second> pps - (2x <per direction packets-per-second>)" - - Aggregate bandwidth: "BANDWIDTH: <aggregate bandwidth in Gigabits per - second> Gbps (untagged)" +From SUT and DUT perspectives, all performance tests involve forwarding +packets between two physical Ethernet ports (10GE, 25GE, 40GE, 100GE). +In most cases both physical ports on SUT are located on the same +NIC. The only exceptions are link bonding and 100GE tests. In the latter +case only one port per NIC can be driven at linerate due to PCIe Gen3 +x16 slot bandwidth limiations. 100GE NICs are not supported in PCIe Gen3 +x8 slots. -- PDR binary search per :rfc:`2544`: +Note that reported DPDK DUT performance results are specific to the SUTs +tested. SUTs with other processors than the ones used in FD.io lab are +likely to yield different results. A good rule of thumb, that can be +applied to estimate DPDK packet thoughput for NIC-to-NIC switching +topology, is to expect the forwarding performance to be proportional to +processor core frequency for the same processor architecture, assuming +processor is the only limiting factor and all other SUT parameters are +equivalent to FD.io CSIT environment. - - Packet rate: "RATE: <aggregate packet rate in packets-per-second> pps (2x - <per direction packets-per-second>)" - - Aggregate bandwidth: "BANDWIDTH: <aggregate bandwidth in Gigabits per - second> Gbps (untagged)" - - Packet loss tolerance: "LOSS_ACCEPTANCE <accepted percentage of packets - lost at PDR rate>"" - -- NDR and PDR are measured for the following L2 frame sizes: - - - IPv4: 64B, 1518B, 9000B. - -All rates are reported from external Traffic Generator perspective. - - -Methodology: Packet Latency ---------------------------- - -TRex Traffic Generator (TG) is used for measuring latency of Testpmd DUTs. -Reported latency values are measured using following methodology: - -- Latency tests are performed at 10%, 50% of discovered NDR rate (non drop rate) - for each NDR throughput test and packet size (except IMIX). -- TG sends dedicated latency streams, one per direction, each at the rate of - 10kpps at the prescribed packet size; these are sent in addition to the main - load streams. -- TG reports min/avg/max latency values per stream direction, hence two sets - of latency values are reported per test case; future release of TRex is - expected to report latency percentiles. -- Reported latency values are aggregate across two SUTs due to three node - topology used for all performance tests; for per SUT latency, reported value - should be divided by two. -- 1usec is the measurement accuracy advertised by TRex TG for the setup used in - FD.io labs used by CSIT project. -- TRex setup introduces an always-on error of about 2*2usec per latency flow - - additonal Tx/Rx interface latency induced by TRex SW writing and reading - packet timestamps on CPU cores without HW acceleration on NICs closer to the - interface line. +Performance Tests Coverage +-------------------------- -Methodology: TRex Traffic Generator Usage ------------------------------------------ +Performance tests measure following metrics for tested DPDK DUT +topologies and configurations: -The `TRex traffic generator <https://wiki.fd.io/view/TRex>`_ is used for all -CSIT performance tests. TRex stateless mode is used to measure NDR and PDR -throughputs using binary search (NDR and PDR discovery tests) and for quick -checks of DUT performance against the reference NDRs (NDR check tests) for -specific configuration. +- Packet Throughput: measured in accordance with :rfc:`2544`, using + FD.io CSIT Multiple Loss Ratio search (MLRsearch), an optimized binary + search algorithm, producing throughput at different Packet Loss Ratio + (PLR) values: -TRex is installed and run on the TG compute node. The typical procedure is: + - Non Drop Rate (NDR): packet throughput at PLR=0%. + - Partial Drop Rate (PDR): packet throughput at PLR=0.5%. -- If the TRex is not already installed on TG, it is installed in the - suite setup phase - see `TRex intallation`_. -- TRex configuration is set in its configuration file - :: +- One-Way Packet Latency: measured at different offered packet loads: - /etc/trex_cfg.yaml + - 100% of discovered NDR throughput. + - 100% of discovered PDR throughput. -- TRex is started in the background mode - :: +- Maximum Receive Rate (MRR): measure packet forwarding rate under the + maximum load offered by traffic generator over a set trial duration, + regardless of packet loss. Maximum load for specified Ethernet frame + size is set to the bi-directional link rate. - $ sh -c 'cd <t-rex-install-dir>/scripts/ && sudo nohup ./t-rex-64 -i -c 7 --iom 0 > /tmp/trex.log 2>&1 &' > /dev/null +|csit-release| includes following performance test suites, listed per NIC type: -- There are traffic streams dynamically prepared for each test, based on traffic - profiles. The traffic is sent and the statistics obtained using - :command:`trex_stl_lib.api.STLClient`. +- **L2IntLoop** - L2 Interface Loop forwarding any Ethernet frames between + two Interfaces. -**Measuring packet loss** +- **IPv4 Routed Forwarding** - L3 IP forwarding of Ethernet frames between + two Interfaces. -- Create an instance of STLClient -- Connect to the client -- Add all streams -- Clear statistics -- Send the traffic for defined time -- Get the statistics +Execution of performance tests takes time, especially the throughput +tests. Due to limited HW testbed resources available within FD.io labs +hosted by :abbr:`LF (Linux Foundation)`, the number of tests for some +NIC models has been limited to few baseline tests. -If there is a warm-up phase required, the traffic is sent also before test and -the statistics are ignored. +Performance Tests Naming +------------------------ -**Measuring latency** +FD.io |csit-release| follows a common structured naming convention for +all performance and system functional tests, introduced in CSIT-17.01. -If measurement of latency is requested, two more packet streams are created (one -for each direction) with TRex flow_stats parameter set to STLFlowLatencyStats. In -that case, returned statistics will also include min/avg/max latency values. +The naming should be intuitive for majority of the tests. Complete +description of FD.io CSIT test naming convention is provided on +:ref:`csit_test_naming`. diff --git a/docs/report/dpdk_performance_tests/packet_latency_graphs/index.rst b/docs/report/dpdk_performance_tests/packet_latency_graphs/index.rst index 014d1095a7..43f557c517 100644 --- a/docs/report/dpdk_performance_tests/packet_latency_graphs/index.rst +++ b/docs/report/dpdk_performance_tests/packet_latency_graphs/index.rst @@ -22,7 +22,7 @@ TGint2-to-SUT2-to-SUT1-to-TGint1. `FD.io test executor dpdk performance job 3n-hsw`_, `FD.io test executor dpdk performance job 3n-skx`_ and `FD.io test executor dpdk performance job 2n-skx`_ with Robot Framework - result files csit-vpp-perf-|srelease|-\*.zip + result files csit-dpdk-perf-|srelease|-\*.zip `archived here <../../_static/archive/>`_. Plotted data set size per test case is equal to the number of job executions presented in this report version: **10**. diff --git a/docs/report/dpdk_performance_tests/packet_throughput_graphs/index.rst b/docs/report/dpdk_performance_tests/packet_throughput_graphs/index.rst index a243922798..cfa75419e6 100644 --- a/docs/report/dpdk_performance_tests/packet_throughput_graphs/index.rst +++ b/docs/report/dpdk_performance_tests/packet_throughput_graphs/index.rst @@ -31,7 +31,7 @@ and their indices. `FD.io test executor dpdk performance job 3n-hsw`_, `FD.io test executor dpdk performance job 3n-skx`_ and `FD.io test executor dpdk performance job 2n-skx`_ with Robot Framework - result files csit-vpp-perf-|srelease|-\*.zip + result files csit-dpdk-perf-|srelease|-\*.zip `archived here <../../_static/archive/>`_. Plotted data set size per test case is equal to the number of job executions presented in this report version: **10**. diff --git a/docs/report/dpdk_performance_tests/test_environment.rst b/docs/report/dpdk_performance_tests/test_environment.rst index 204792015f..7de29ee332 100644 --- a/docs/report/dpdk_performance_tests/test_environment.rst +++ b/docs/report/dpdk_performance_tests/test_environment.rst @@ -1,8 +1,8 @@ -.. include:: ../vpp_performance_tests/test_environment_intro.rst +.. include:: ../introduction/test_environment_intro.rst -.. include:: ../vpp_performance_tests/test_environment_sut_conf_1.rst +.. include:: ../introduction/test_environment_sut_conf_1.rst -.. include:: ../vpp_performance_tests/test_environment_sut_conf_3.rst +.. include:: ../introduction/test_environment_sut_conf_3.rst DUT Configuration - DPDK @@ -20,49 +20,23 @@ DUT Configuration - DPDK **Testpmd Startup Configuration** -Testpmd startup configuration changes per test case with different settings for CPU -cores, rx-queues. Startup config is aligned with applied test case tag: - -Tagged by **1T1C** - -.. code-block:: bash - - testpmd -c 0x3 -n 4 -- --numa --nb-ports=2 --portmask=0x3 --nb-cores=1 --max-pkt-len=9000 --txqflags=0 --forward-mode=io --rxq=1 --txq=1 --burst=64 --rxd=1024 --txd=1024 --disable-link-check --auto-start - -Tagged by **2T2C** - -.. code-block:: bash - - testpmd -c 0x403 -n 4 -- --numa --nb-ports=2 --portmask=0x3 --nb-cores=2 --max-pkt-len=9000 --txqflags=0 --forward-mode=io --rxq=1 --txq=1 --burst=64 --rxd=1024 --txd=1024 --disable-link-check --auto-start - -Tagged by **4T4C** +Testpmd startup configuration changes per test case with different settings for +`$$CORES`, `$$RXQ` and max-pkt-len parameter if test is sending jumbo frames. +Startup command template: .. code-block:: bash - testpmd -c 0xc07 -n 4 -- --numa --nb-ports=2 --portmask=0x3 --nb-cores=4 --max-pkt-len=9000 --txqflags=0 --forward-mode=io --rxq=2 --txq=2 --burst=64 --rxd=1024 --txd=1024 --disable-link-check --auto-start + testpmd -c $$CORE_MASK -n 4 -- --numa --nb-ports=2 --portmask=0x3 --nb-cores=$$CORES --max-pkt-len=9000 --txqflags=0 --forward-mode=io --rxq=$$RXQ --txq=$$TXQ --burst=64 --rxd=1024 --txd=1024 --disable-link-check --auto-start **L3FWD Startup Configuration** -L3FWD startup configuration changes per test case with different settings for CPU -cores, rx-queues. Startup config is aligned with applied test case tag: - -Tagged by **1T1C** - -.. code-block:: bash - - l3fwd -l 1 -n 4 -- -P -L -p 0x3 --config='${port_config}' --enable-jumbo --max-pkt-len=9000 --eth-dest=0,${adj_mac0} --eth-dest=1,${adj_mac1} --parse-ptype - -Tagged by **2T2C** - -.. code-block:: bash - - l3fwd -l 1,2 -n 4 -- -P -L -p 0x3 --config='${port_config}' --enable-jumbo --max-pkt-len=9000 --eth-dest=0,${adj_mac0} --eth-dest=1,${adj_mac1} --parse-ptype - -Tagged by **4T4C** +L3FWD startup configuration changes per test case with different settings for +`$$CORES` and enable-jumbo parameter if test is sending jumbo frames. +Startup command template: .. code-block:: bash - l3fwd -l 1,2,3,4 -n 4 -- -P -L -p 0x3 --config='${port_config}' --enable-jumbo --max-pkt-len=9000 --eth-dest=0,${adj_mac0} --eth-dest=1,${adj_mac1} --parse-ptype + l3fwd -l $$CORE_LIST -n 4 -- -P -L -p 0x3 --config='${port_config}' --enable-jumbo --max-pkt-len=9000 --eth-dest=0,${adj_mac0} --eth-dest=1,${adj_mac1} --parse-ptype -.. include:: ../vpp_performance_tests/test_environment_tg.rst +.. include:: ../introduction/test_environment_tg.rst |